Diabetes mellitus (DM) is an increasing threat to human health. Long-term exposure to DM can lead to devastating macro- and microvascular complications, including cardiovascular disease, blindness, amputation, and kidney failure via diabetic nephropathy (DN). However, not all diabetics develop complications at the same rate, and the causes of this variation in risk are largely unknown. Using a rich set of genome-wide gene expression profiles of 1,240 Mexican Americans from more than 40 extended families, we have identified 22 genes that exhibit highly significant genotype x duration of DM interaction effects. While many genes are differentially expressed in DM, this subset represents those genes with the greatest genetic variability in the response to disease exposure. The subset is enriched for genes related to oxidative stress and inflammation, including SMAD1, a putative early biomarker of DN. The subset is connected to thousands of other genes through both known pathways and via genetic correlation in the sample. In this study, we will combine genetic, transcriptomic, and bioinformatic data to identify gene networks with variable response to DM exposure, to develop hypotheses of causation and to search for new gene variants that may underlie individual variation in risk for complications. We will then test these hypotheses in the context of DN, using kidney function phenotypes measured in 858 of these same individuals ~10 years after acquisition of the expression phenotypes, as well as in 2,282 Mexican Americans from replicate samples recruited on probands with established DM and/or DN.